Optimization of ligand-binding assay in a QbD environment. Use of Bayesian non-linear regression to set up probability profile as quality response

For ligan-binding assay, different parameters are controlled in routine. In a development phase, we will show how design of experiment can be used to select these parameters to obtain reliable results in the future use of the ELISA test. Uncertainties of measurements and models should be taken into account in a predictive manner. Bayesian modelling and MCMC simulations are well suited for this purpose and are flexible enough to include random effects on the parameters on non-linear model. We will focus on 4PL regression as it is encountered in most applications. We will also show that deriving a quality response such as a precision profile is clear and direct in the Bayesian setting. The approach is fully compliant with QbD guidelines and Design Space definition.